Moore's law, the doubling of the number of transistors on a chip along with the halving of their cost every 18 months has reached its end (at least for the cost part).
A similar pattern has been true over the last 35 years (Moore's law lasted 50 years) for data transmission over optical fibre. Here the doubling referred to the fibre capacity (the speed is fixed: 2/3 of the speed of light in a vacuum) and to the decrease of cost per bit (a consequence of the basically constant price of fibre and termination points).
This pattern can be associated to Donal Keck, the co-inventor of low-loss optical fibre.
We have seen a 10 million fold improvement in capacity on a single fibre since 1980 (a doubling every 2 years, approx.). In the labs with the very latest technologies one can carry one million billion bits per second on a single fibre.
This is simply amazing: the good news is that after a few years of overcapacity in the first part of the last decade now the explosion of streaming in HD and UHD is making use of this capacity. The bad news is that it looks like we have run short of ideas to further increase the capacity of a fibre (and the Shannon limit is just in front of us).
Clearly, we can deploy more fibres but this is increasing the cost per bit, hence, like in the case of the Moore's law we still have ways to increase performance but at a cost! And this happens after so many years of cost decrease that managed to create the perception that communication and processing is basically free.